Evidence is also presented that the BOGUAY strain may possess heterotrophic as well as autotrophic carbon uptake capabilities, and at least two energy-producing electron transport chains. A single filament collected from core 4489-10 (Fig. 1) from RV Atlantis/HOV Alvin cruise AT15-40 (13 December

2008) at the UNC Gradient Mat Epacadostat site in Guaymas Basin, Gulf of California (latitude 27° 00.450300′ N, longitude 111° 24.532320′ W, depth 2001 m) was cleaned of epibionts; its DNA amplified, tested for genetic purity, sequenced, and annotated; and the genome sequence checked for completeness, as previously described ( MacGregor et al., 2013a). A total of 99.3% of the sequence was assembled into 822 contigs, suggesting good coverage was achieved. A total of 4.7 Mb of sequence was recovered with 80% of the sequence forming large (≥ 15 kb) contigs. Throughout this paper,

annotated sequences will be referred to by 5-digit contig and 4-digit open reading frame (ORF) numbers, e.g., 00024_0691. Additional sequence analysis was carried out using a combination of the JCVI-supplied annotation, the IMG/ER ( Markowitz et al., 2009) and RAST ( Aziz et al., 2008) platforms, and BLASTN, BLASTX, and BLASTP, PSIBLAST, and DELTABLAST searches of the GenBank nr databases. Nucleic acid and amino acid sequence alignments were performed in MEGA5 ( Tamura Selleck PD 332991 et al., 2011) using MUSCLE ( Edgar, 2004) or with the NCBI COBALT aligner ( Papadopoulos and Agarwala, 2007) and small adjustments made manually. Maximum-likelihood

phylogenies were inferred in ARB ( Ludwig et al., 2004) with RAxML rapid bootstrapping ( Stamatakis, 2006) using a random initial tree, the PROTMIX SPTLC1 rate distribution and WAG amino acid substitution models (unless a different substitution model was identified as most likely in a Bayesian run), empirical amino acid frequencies, and branch optimization. Bayesian phylogenies were inferred in MrBayes 3.2 ( Ronquist et al., 2012), run as two sets of four Markov chain Monte Carlo runs until these converged. A mixed prior amino acid substitution model was chosen. In nearly all cases, the WAG model had a posterior probability of 1.000 (see figure legends for exceptions); if not, RAxML was rerun with the model identified. Bayesian trees were displayed with FigTree 1.4 (http://tree.bio.ed.ac.uk/software/figtree/). For the phylogenetic analyses shown here, all relatively full-length BLASTP matches in the NCBI nr database up to a total of 100 were first used to build bootstrapped neighbor-joining trees in ARB. From these, approximately 50 of the more closely related sequences plus 3–5 outgroup sequences were selected for RAxML analysis. Sequences from the final RAxML tree were then exported to MrBayes.